diff --git a/testing/web-platform/meta/MANIFEST.json b/testing/web-platform/meta/MANIFEST.json
index 2326f79c5ec0..59a206bfa388 100644
--- a/testing/web-platform/meta/MANIFEST.json
+++ b/testing/web-platform/meta/MANIFEST.json
@@ -615453,7 +615453,7 @@
    "support"
   ],
   "url/README.md": [
-   "17ffb158bda35ef8b55f8f7f5f084df2e0651bfc",
+   "37bbc0fb70e865c45ff369a434de1a4a0a2bbc3e",
    "support"
   ],
   "url/a-element-origin-xhtml.xhtml": [
diff --git a/testing/web-platform/tests/url/README.md b/testing/web-platform/tests/url/README.md
index 34c35dc4a352..667951d680fc 100644
--- a/testing/web-platform/tests/url/README.md
+++ b/testing/web-platform/tests/url/README.md
@@ -23,7 +23,7 @@ The keys for each test case are:
 
 In addition to testing that parsing `input` against `base` gives the result, a test harness for the
 `URL` constructor (or similar APIs) should additionally test the following pattern: if `failure` is
-true, parsing `about:blank` against `base` must give failure. This tests that the logic for
+true, parsing `about:blank` against `input` must give failure. This tests that the logic for
 converting base URLs into strings properly fails the whole parsing algorithm if the base URL cannot
 be parsed.